United States of America: Selected Issues

This Selected Issues paper on the United States examines the effect of the structure of the mortgage market on real housing activity and housing prices. The market-based financial structure has reduced the volatility of mortgage lending. Changes in the structure of the mortgage market have coincided with lower volatility of real housing activity. Regional income growth and unemployment rates have statistically significant and correct signed effects on housing prices. Tests of the relative importance of mortgage market structure and macroeconomic variables suggest an important effect from the financial structure.

Abstract

This Selected Issues paper on the United States examines the effect of the structure of the mortgage market on real housing activity and housing prices. The market-based financial structure has reduced the volatility of mortgage lending. Changes in the structure of the mortgage market have coincided with lower volatility of real housing activity. Regional income growth and unemployment rates have statistically significant and correct signed effects on housing prices. Tests of the relative importance of mortgage market structure and macroeconomic variables suggest an important effect from the financial structure.

PART II: EXTERNAL LINKAGES

III. Why Has the U.S. Trade Balance Widened So Fast?14

A. Introduction and Summary

1. The rapid decline of the U.S. trade balance in recent years presents something of a puzzle. Notwithstanding a 15 percent real depreciation of the dollar between 2002 and 2004, the U.S. current account deficit increased from 4½ percent of GDP to 5¾ percent of GDP, mainly driven by a widening trade deficit. Many forecasters have been surprised that the weaker dollar appears to have induced neither a slowdown in real imports nor an acceleration of real exports.

2. Explanations for the rising trade deficit often focus on differences in growth rates and trade elasticities between the United States and its trade partners:15

  • The large gap between U.S. exports and import levels, as well as rapid growth in the United States compared with its trading partners, mean that the trade deficit would widen even if income elasticities of exports and imports were similar across countries (e.g., Greenspan, 2003).

  • The income elasticity of U.S. imports is also typically estimated to be higher than the foreign income elasticities of U.S. exports (Houthakker and Magee, 1969). While the existence and implications of differential income elasticities are still a matter of debate, they would imply a widening trade deficit even under similar growth rates.16

3. More recently, attention has focused on the growing penetration of U.S. markets by Chinese imports and a—possibly related—decline in exchange rate pass-through to import prices.17 This reflects the fact that traditionally estimated growth and elasticity differentials may not be sufficient to explain the recent deterioration in the trade balance. It also raises the question whether the real effective exchange rate remains an accurate gauge of the overall competitiveness of the U.S. economy.

4. This paper finds some link between shifts in the geographical composition of imports and the decline in pass-through. Using cross-country information, we find that changes in a country’s trade structure are associated with lower pass-through. These results suggest that rapid changes in the trade structure may have reduced the ability of trade models to explain recent developments in the trade balance.

5. The paper also finds that recent developments in the trade balance cannot be fully explained by differential income elasticities. Reflecting findings of a declining pass-through, the staff’s trade model is modified to include relative prices of imports and exports—rather than the effective exchange rate—in the trade volume equations. Two variants of this new model are used to explore whether accounting for possible asymmetries in income elasticities—the Houthakker-Magee effect—can help explain the deterioration in the trade balance. Despite improvements in fit, the models cannot fully account for the widening trade deficit, partly due to the recent unimpressive performance of U.S. exports.

B. Pass-Through and Shifts in the Trade Structure

6. The relationship between exchange rate changes and U.S. import prices has loosened considerably in recent years. Despite a significant nominal and real effective depreciation of the dollar since 2002, the relative price of imports remained broadly stable (Figure 1), suggesting a lower pass-through coefficient.

Figure 1.
Figure 1.

Relative import prices have not risen as fast as the dollar has depreciated.

Citation: IMF Staff Country Reports 2005, 258; 10.5089/9781451839647.002.A003

Sources: BEA; and Board of Governors, Federal Reserve.

7. A simple equation was used to analyze how exchange rate pass-through has changed over time. Staff’s standard trade model includes the following import price equation:

ΔpM  =β0+β1Δp1M+β2Δp1*+β3Δp-1Y+β4ΔE(1)β5(p1Mp1Yβ6e1β7t)+ε

where all variables are in logs (see Annex for variable definitions and sources) and lags are denoted by the subscript -1. Over the long run, import prices move together with domestic prices (pY) and the real effective exchange rate (e), as well as a time trend. Over the short run, changes in import prices are influenced by its own recent behavior, past changes in foreign and domestic prices (Δ p* and ΔpY); contemporaneous changes in the nominal effective exchange rate (ΔE); and a measure of the distance from the long-term equilibrium (an “error correction term”). Similar equations are estimated for subindexes, such as import prices excluding commodity prices.

8. The estimated coefficients suggest that pass-through has weakened in recent years (Figure 2). To capture possible changes in the strength of pass-through—defined as the short-run coefficient on the nominal exchange rate (β4))—we perform rolling estimations (with a 10-year window) on quarterly data over the 1984-2004 period. The results suggest that this coefficient has dropped from about -0.3 in 1984-1994 to around -0.1 more recently.18 Much of the decline in the strength of pass-through appears to have happened since 2000, driven by changes in the sensitivity of non-commodity goods to exchange rate movements (Figure 3).19

Figure 2.
Figure 2.

Estimated short-term pass-through coefficients

(all commodities)

Citation: IMF Staff Country Reports 2005, 258; 10.5089/9781451839647.002.A003

Figure 3.
Figure 3.

Estimated short-run pass-through coefficients

Citation: IMF Staff Country Reports 2005, 258; 10.5089/9781451839647.002.A003

9. The results are robust to using import weights instead of overall trade weights in the construction of the effective exchange rate. Obtaining relevant pass-through coefficients depends on using the most appropriate exchange rate concept, particularly in periods when the trade structure is changing rapidly. For example, if the composition of imports changes relatively faster, using a broad trade-weighted effective exchange rate measure could contribute to underestimating the strength of exchange rate pass-through. The model was therefore re-estimated with an import-weighted exchange rate, with only minor differences in the results.20

Import prices and globalization

10. The reasons for the decline in pass-through are not well understood, but a number of explanations have been proposed, several of them related to globalization:

  • Substitution effects arising from the growing presence of low-cost producers in world markets may dampen import prices. Given China’s peg to the dollar, a shift in U.S. import composition towards Chinese goods in recent years may be particularly relevant in dampening U.S. pass-through (Figure 4).

  • High productivity growth in the export sector of newly-integrating countries may help contain or even reduce their export prices.

  • Potential competition from emerging low-cost producers may decrease other producers’ willingness to increase prices if their currencies appreciate.

  • Foreign exporters may increasingly “price to market” by absorbing exchange rate changes into profit margins, reflecting competitive pressures from more open trade as well as cyclical factors.

  • Trade composition may also play a role. If the share of goods with exchange rate sensitive prices—e.g., commodities—in total imports fall, aggregate pass-through may decline. An increase in the share of goods with high productivity growth—e.g., high-tech goods—may also depress pass-through.

Figure 4.
Figure 4.

U.S. Effective Exchange Rate Weights, 1995 and 2002

Citation: IMF Staff Country Reports 2005, 258; 10.5089/9781451839647.002.A003

11. This paper confirms that import price inflation tended to be lower in sectors where China’s share in imports—an indicator of the impact of of low-cost producers—increased more. Following Kamin and others (2004) and Marazzi and others (2005), we use 5-digit end-use category import price and trade share data. Although our sample is constrained by data availability to 49 categories, it covers about 50 percent of total U.S. imports, and about 75 percent of U.S. imports from China.21 We consider the change in China’s share in imports as a simple indicator of a sector’s exposure to globalization. Figure 5 (see previous page) shows that this indicator varies significantly across sectors, and that some sectors experienced a significant increase over the past decade.

Figure 5.
Figure 5.

China’s import share of selected end-use goods

In percent of all U.S. imports by sector

Citation: IMF Staff Country Reports 2005, 258; 10.5089/9781451839647.002.A003

Source: Census Bureau.

12. Tests using sectoral averages suggest a negative correlation of Chinese import penetration and import price inflation, as found by Kamin and others (2004). Running a simple OLS regression confirms that sectors with a large increase in China’s import share tended to experience lower import price inflation (Table 1 and Figure 6). Comparing data before and after 2000 also indicates that this effect may have become stronger in recent years.

Table 1.

OLS Regression Results: Determinants of Average Import Price Inflation Across Sectors

Dependent variable: Average import price inflation, 1993-2000

article image
Sources: Bureau of Census; Haver Analytics; Kamin and others (2004); and IMF staff calculations.

Average over shorter periods for sectors where price data start later than 1985.

Figure 6.
Figure 6.

Import Prices and China’s Import Share

Citation: IMF Staff Country Reports 2005, 258; 10.5089/9781451839647.002.A003

Source: Census Bureau.

13. There is also some evidence that Chinese import penetration may be associated with reduced volatility of import inflation. In sectors where China’s import share was large, or where China’s import share increased rapidly, the standard deviation of import price changes (measured relative to the GDP deflator) tended to be lower (Table 2).22

Table 2.

OLS Regression Results: Determinants of the Standard Deviation of Relative Import Price Inflation Across Sectors

Dependent variable: Standard deviation of changes in relative import prices, 1995-2004 1

article image
Sources: Bureau of Census; Haver Analytics; and IMF staff calculations.

Import prices relative to GDP deflator.

Standard deviation over shorter period for sectors where price data start after 1985.

14. Earlier work has also identified a link between Chinese import penetration and pass-through. Marazzi and others (2005) postulated that, if globalization was associated with changes in import pricing, the decline in pass-through would be especially pronounced in sectors that became more exposed to imports from low-cost countries. They estimated pass-through coefficients by sector over two 10-year periods, 1985-94 and 1995-04. By correlating the change in sector-specific pass-through with the change in China’s import share, they found a statistically significant relationship between China’s import share and pass-through.

15. We test a similar relationship between Chinese import penetration and pass-through coefficients. Applying the two-stage methodology of Marazzi and others (2005), sector-specific versions of equation (1) were estimated to obtain sectoral pass-through coefficients for 1985-94 and 1995-2004.23 The link between changes in pass-through and increases in import penetration from China was tested by estimating the following cross-sectional regression:

β4,i9504=γ0+γ1β4,i8594+γ2ΔChinai+ϵi(2)

where β4, i indicates the two sets of sectoral pass-through coefficients, and ΔChina is the change in the share of Chinese imports. Based on the earlier results reported by Marazzi and others, the coefficient γ2 would be expected to be significant and positive.

16. The results do not suggest that the decline in pass-through is strongly related to Chinese import penetration. The estimated coefficient has the expected sign but is not statistically significant. Figure 7, which depicts the data underlying equation (2), illustrates that pass-through declined in most sectors (the change in the pass-through coefficient is positive). But it also shows that some sectors have seen little change in pass-through despite a large increase in Chinese imports. Although this picture is similar to the one reported in the Marazzi study, the result may differ both because the sectoral coverage of the sample is smaller, and because the underlying import price equations used to estimate sectoral pass-through are different. Also, this paper did not have access to a full set of sectoral import weights, requiring the use of proxies for sectoral exchange rates and foreign exporters’ prices (see Appendix).

Figure 7.
Figure 7.

Passthrough and China’s Share in U.S. Imports

Citation: IMF Staff Country Reports 2005, 258; 10.5089/9781451839647.002.A003

Source: Census Bureau.
Pass-through across countries

17. We use cross-country data to test whether shifts in the trade structure are associated with a change in pass-through. The methodology of Marazzi and others (2005) is applied to relate changes in pass-through coefficients in a set of advanced economies to two different measures of the changing trade structure. One of these measures is—like in the previous section—the change in the share of imports from China. The other measure is an index of structural change based on the import shares (si) of multiple trading partners:

12Σi|si93si04|(3)

18. The results are suggestive of a link between falling pass-through and shifts in the structure of imports, although no direct “China effect” is observable (Table 3). While the coefficient on the change in Chinese import share is both wrongly signed and insignificant, the broader indicator of structural change has the expected and statistically significant impact. At face value, this finding would seem to suggest that pass-through analysis should not be limited to the impact of China alone and other indicators of changes in the trade structure may be usefully considered to capture the possible effect of globalization on exchange rate pass-through.

Table 3.

OLS Regression Results: Pass-through and Indicators of Globalization

Dependent variable: Pass-through coefficient, 1995-2004

article image
Sources: Direction of Trade Statistics; International Finance Statistics; OECD Analytical Database; and IMF staff calculations

C. Revisiting the U.S. Goods Trade Model

19. In view of the decline in pass-through, and considering the difficulties in predicting recent U.S. trade developments, this paper explores room for further improving the performance of a standard empirical trade model. This section first explores if using relative prices—rather than real effective exchange rates—helps to better track the recent deterioration of the trade balance. The analysis then seeks to determine whether a possible asymmetry in the income elasticity of U.S. exports and imports—the Houthakker-Magee effect—has played an important role in accounting for the widening trade deficit.

20. The starting point for the analysis is a multi-equation trade model describing import and export prices and volumes. As reported in last year’s Staff Report, this baseline model allows for long run and short-run dynamics through the introduction of an error correction mechanism. The model also allowed for productivity-related effects that can boost both a country’s domestic growth and the demand for exports from that country (“supply effects”).

21. The baseline model was revised in two ways, partly reflecting the results of the previous section. First, as mentioned, we introduce an alternative measure of the real exchange rate. Specifically, in the import equation the ratio of import prices over domestic prices is substituted for the real effective exchange rate, which should improve robustness vis-à-vis changes in pass-through. Second, using this basic model, we then compare the performance of a specification with both demand and supply effects to a Houthakker-Magee model where only demand in the importing country enters the equation. In this way, it is possible to gauge whether differences in income elasticities are partly responsible for the recent evolution of the trade balance.

22. In formal terms, the basic model is provided by the following import equation:

Δmt=c+αyΔyt+αrpΔrptm+γ(mtβyytβy*yt*βrprptmβrpfrptmfβrpnfrptmnf)(4)

The specification is used correspondingly for the export equation. Variables are defined as follows (all in logs):

m U.S. goods import volumes.

y U.S. real GDP.

rpm Relative price of imports (nominal import prices over GDP deflator).

y* Foreign real GDP.

rpmf Nominal fuel import prices over GDP deflator.

rpmnf Nominal non-fuel commodity import prices over GDP deflator.

23. We test the performance of this model under two different restrictions for long-run income elasticities:

  • The supply-demand (SD) model imposes: βy = βy*.24

  • For the demand-only (DO) model, we set: βy* = 0.

24. The two variations of the model fit the historical data equally well, although estimates for the long-run income elasticities differ sharply. Table 4 reports coefficient estimates for import and export volumes under both specifications. Although standard measures of fit are almost identical across models, the long-run income elasticity of both imports and exports is substantially higher for the DO model. As for the Houthakker-Magee effect, the estimated income elasticities for imports and exports are almost identical in the SD model, but differ more substantially in the DO model. This illustrates the difficulties in establishing the empirical importance of the Houthakker-Magee effect.

Table 4.

Coefficient Estimates with the New Models

(sample 1980:4 2004:2)

Import of Goods Volumes

article image
Souces: Fund staff estimates.
In-sample forecasts

25. The revised specification improves the overall fit of the staff’s trade model(Figure 8). The two new models perform quite similarly, at least over a short forecast horizon. In particular, the use of relative prices in the volume equations makes the model better suited to explain the strength of U.S. imports. This would suggest that the trade-weighted real effective exchange rate may overestimate the degree to which dollar depreciation has improved the competitiveness of domestic producers. It is also notable that the Houthakker-Magee effect appears to play a relatively minor role, despite a substantial growth differential between the U.S. and most trade partners. This result would tentatively suggest that while differential growth rates may have been a factor behind the recent decline in the trade balance, the role of asymmetries in income elasticities has remained limited.

Figure 8.
Figure 8.

U.S. Trade Balance: In-sample forecasts and actual values

Citation: IMF Staff Country Reports 2005, 258; 10.5089/9781451839647.002.A003

26. However, despite improvements over last year’s specification, the trade model still fails to explain much of the decline in the trade balance since early 2002. This owes in large part to the weak performance of the export equations, which continue to overpredict export volumes. We checked whether a different measure for income growth in U.S. export markets could improve the model fit, given that aggregate trade weights could be biased by large changes in the composition of imports without corresponding changes in the composition of exports. However, an export-weighted measure of foreign income with time-varying weights yielded no improvement in fit.25 The following section therefore explores whether export relationships have suffered a structural break, or whether foreign demand has become more sensitive to fluctuations in foreign income.

Stability analysis and the responsiveness to relative prices

27. Evolving trade patterns may have precipitated changes in the short-run elasticity of both exports and imports to movements in relative prices.26 As already discussed, the penetration of China and other low-cost exporters into import markets in industrialized countries may have induced changes in competitiveness and pricing practices. Even after accounting for structural changes in the price equations, it is still possible that the response of volumes to (relative) import and export price movements have been similarly affected by greater trade integration.

28. Rolling regressions indeed suggest that goods import volumes have become less responsive not only to real effective exchange rates but also relative prices. The short-run coefficients for relative import and export prices are shown in Figure 9, based on a moving window of 60 observations. The results suggest a dramatic drop in the sensitivity of imports to relative prices in recent years, to the point that the coefficient is no longer significantly different from zero. Estimating the equations for commodities and non-commodities separately reveals that this is due to a lower relative price sensitivity for non-commodities, for which competition may have increased more strongly.

Figure 9.
Figure 9.

Export and Import Volume Relative Price Elasticities

(rolling 60 quarter correlation with +/- 1.65 standard deviation bands)

Citation: IMF Staff Country Reports 2005, 258; 10.5089/9781451839647.002.A003

Source: IMF staff estimates.

29. The opposite result holds for U.S. goods exports, however. Overall exports of goods, in particular non-commodities, appear to have become somewhat more sensitive to short-term changes in relative prices. This finding deepens the puzzle over the recent U.S. export performance, as it would imply a larger boost to export volumes following the dollar’s depreciation.

30. Stability analysis also reveals increased responsiveness of exports to foreign income. Rolling regressions show a rise in the coefficient measuring the short-run export response to foreign income growth. This would suggest that the sensitivity of the U.S. current account balance to foreign growth has increased in recent years, helping to explain some of the weakness observed in recent years.

31. Increased openness to trade and foreign competition might help rationalize these changes. Analyzing the possible causes for the observed change in short-run export elasticities vis-à-vis relative prices and income lies beyond the current analysis. However, we surmise that with greater openness to trade, the rise in the number of suppliers for a given good may have increased both competition and the sensitivity of exports to relative price movements. Conversely, this could explain the lower responsiveness of imports to relative price fluctuations, as the presence of a greater number of competitors within one market facilitates expenditure switching across similar goods by consumers. It is also plausible that increased openness to trade may have magnified the transmission of international shocks, particularly through trade channels.27

Out–of-sample forecasts

32. Comparing the out-of-sample prediction of the new models sheds some light on the role of income-elasticity differentials for the evolution of the trade balance. Point estimates presented earlier indicate a slight difference in the long-run income elasticities between exports and imports, as well as a larger sensitivity to income changes for both equations in the DO specification compared to the SD version. As the forecasting performance of both models is fairly similar for short horizons (see above), assessing the possible role of the Houthakker-Magee effect and the importance of supply versus demand effects requires comparing forecasts over a longer horizon.

33. Discrepancies in forecast paths suggest some limited role for elasticity differentials in accounting for the widening trade deficit. Figure 10 presents the out-of-sample forecasts obtained through 2010Q4 with the new models using April 2005 World Economic Outlook data. The improvement in the trade deficit suggested by both models arises from the lagged effects of the U.S. exchange rate depreciation combined with the assumption of an acceleration in foreign output growth rates that entails a convergence towards U.S. levels. While the DO model—which incorporates a Houthakker-Magee effect—shows a slightly larger trade deficit over time, the difference between the two models amounts to a surprisingly modest ½ percentage point of GDP over 5 years.

Figure 10.
Figure 10.

U.S. Trade Balance: Out-of-sample forecasts

Citation: IMF Staff Country Reports 2005, 258; 10.5089/9781451839647.002.A003

Source: IMF staff estimates.

34. While the two models’ forecasts show a similar sensitivity to exchange rate shocks, they react differently to changes in growth differentials. A depreciation of the dollar and a narrowing of growth differentials between the U.S. and its major trade partners are often mentioned as factors in a gradual adjustment of the current account deficit. The two models would predict similar effect from exchange rate changes. However, the effects of a sustained acceleration of growth abroad compared with the baseline differ. Due to the higher long-run income sensitivity of exports estimated in the DO model, the predicted increase is much larger compared to the model with supply effects.

APPENDIX: Data and Sources

Data for estimating aggregate pass-through equations

1. Table 5 describes the variables used for estimating equation (1). In alternative specifications, price indices for non-oil imports and noncommodity imports were used as the left-hand side variable. Some specifications included oil and commodity prices as explanatory variables (source: WEO).

Table 5.

Data Definitions and Sources for Aggregate Pass-Through Equations

article image

Data for sectoral estimates

2. Sectoral pass-through equations were estimated based on data described in Table 6.

Table 6.

Data Definitions and Sources for Sectoral Pass-Through Equations

article image

3. We experimented with two sets of proxies for the (nominal and real) exchange rate and foreign prices. Ideally, these variables should be sector-specific. However, this paper did not have access to sectoral data on the geographical composition of imports for the full sample period, which required constructing proxies:

  • The first set of proxies is simply the “broad” exchange rate and the aggregate foreign price variable that we used in the previous subsection. The broad exchange rate is constructed with time-varying weights based on the overall trade structure, i.e., it ignores cross-sectoral differences which, as Figure 5 illustrates, can be substantial.

  • The other set of proxies includes an effective exchange rate and a foreign price variable based on constant 2004 sector-specific weights. This accounts better for differences in trade structure across sectors (at least in 2004), but ignores the changing importance of different trading partners over time. Again, Figure 5 illustrates that this omission can be important.

4. The pass-through coefficients estimated using the “broad” and “sectoral” explanatory variables were found to be quite closely related. In line with intuition, sectoral-based estimates are significantly smaller, particularly in the earlier period; and the broad-based coefficient estimates vary across a broader range for the various sectors.

5. China’s share in imports by end-use category was calculated based on Bureau of Census data (for 2000-04). For 1993, shares reported by Kamin and others (2004) were used.

Data for country estimates

6. Data cover 17 advanced economies: Australia, Austria, Canada, Denmark, Finland, France, Germany, Italy, Japan, Netherlands, New Zealand, Norway, Spain, Sweden, Switzerland, United Kingdom, United States.

7. Import prices are overall import deflators from the OECD Analytical Database. Domestic prices are represented by GDP deflators (from the OECD Analytical Database), while foreign prices are import-weighted CPIs of trading partners (from the WEO). Nominal and real effective exchange rates are from the International Financial Statistics.

8. Structural change variables—the change in China’s share in total imports and the indicator of change in the trade structure—are constructed based on information from the Direction of Trade Statistics. The structural change indicator is based on the following breakdown of total imports:

  • Industrial countries: Canada, Australia, Japan, New Zealand, Austria, Belgium (Belgium-Luxembourg), Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Italy, Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, United Kingdom, United States

  • Africa

  • Asia: China, P.R., India, Indonesia, Korea, Malaysia, Other

  • Emerging Europe

  • Middle East

  • Western Hemisphere: Argentina, Brazil, Mexico, Other

  • Other

References

  • Bernanke, B., 2005, “The Global Saving Glut and the U.S. Current Account Deficit,” Sandridge Lecture delivered to the Virginia Association of Economists, Richmond (March 10). Available on the Internet at http://www.federalreserve.gov/boarddocs/speeches/2005/200503102/default.htm.

    • Search Google Scholar
    • Export Citation
  • Blanchard, O., P. Giavazzi, and F. Sa, 2005, “The U.S. Current Account and the Dollar,” CEPR Working Paper 4888 (London: Center for Economic Policy Research).

    • Search Google Scholar
    • Export Citation
  • Chinn, M. 2004, “Incomes, Exchange Rates and the U.S. Trade Deficit, Once Again,” International Finance, Vol. 7, pp. 45169.

  • Chinn, M., 2005, “Still Doomed to Deficits: An Update on US Trade Elasticities” (unpublished; University of Wisconsin).

  • Ferguson, R., 2005, “U.S. Current Account Deficit: Causes and Consequences,” Remarks to the Economics Club of the University of North Carolina at Chapel Hill (April 20). Available on the Internet at http://www.federalreserve.gov/boarddocs/speeches/2005/20050420/default.htm.

    • Search Google Scholar
    • Export Citation
  • Greenspan, A, 2005, “Current Account,” Speech delivered at the Advancing Enterprise Conference, London (February 4). Available on the Internet at http://www.federalreserve.gov/boarddocs/speeches/2005/20050204/default.htm.

    • Search Google Scholar
    • Export Citation
  • Hooper, P, K. Johnson, and J. Marquez, 2000, “Trade Elasticities for the G-7 Countries,” Princeton Studies in International Economics No. 87 (Princeton University).

    • Search Google Scholar
    • Export Citation
  • Houthakker, H., and S. Magee, 1969, “Income and Price Elasticities in World Trade,” Review of Economics and Statistics, Vol. 51, pp. 11125.

    • Search Google Scholar
    • Export Citation
  • Kamin, S., M. Marazzi, and J. Schindler, 2004, “Is China ‘Exporting Deflation’?International Finance Discussion Paper No. 791 (Washington: Board of Governors of the Federal Reserve).

    • Search Google Scholar
    • Export Citation
  • Mann, C., 2002, “Perspectives on the U.S. Current Account Deficit and Sustainability”, Journal of Economic Perspectives, Vol. 16, pp. 13152.

    • Search Google Scholar
    • Export Citation
  • Marazzi, M., N. Sheets, R. Vigfusson, J. Faust, J. Gagnon, J. Marquez, R. Martin, T. Reeve, and J. Rogers, 2005, “Exchange Rate Pass-Through to U.S. Import Prices: Some New Evidence,” International Finance Discussion Papers No. 833 (Washington: Board of Governors of the Federal Reserve).

    • Search Google Scholar
    • Export Citation
  • Marquez, J., 2000, “The Puzzling Income Elasticity of U.S. Imports” (unpublished; Washington: Board of Governors of the Federal Reserve).

    • Search Google Scholar
    • Export Citation
  • Obstfeld, M., and K. Rogoff, 2004, “The Unsustainable Current Account Position Revisited,” NBER Working Paper 10869 (Cambridge: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation
  • Roubini, N., and B. Setser, 2004, “The U.S. as a Net Debtor: The Sustainability of the U.S. External Imbalances” (unpublished; New York: Stern School of Business).

    • Search Google Scholar
    • Export Citation
  • Rubin, R., P. Orszag, and A. Sinai, 2004, “Sustained Budget Deficits: Longer-Run U.S. Economic Performance and the Risk of Financial Disarray,” Paper presented at the Allied Social Science Association Annual Meetings (San Diego).

    • Search Google Scholar
    • Export Citation
14

Prepared by Alejandro Justiniano and Kornélia Krajnyák.

15

Mann (2003) notes that there are a number of ways of analyzing the external sector—trade flows, savings and investment relationships, or the financing through the capital account. The different angles serve to highlight particular aspects of the external sector. This paper focuses on the trade dimension. Proposed explanations of for the U.S. current account deficit also include other factors, such as the role of technology and savings opportunities for foreign investors (Ferguson, 2003), an international “savings glut” (Bernanke, 2005), or the U.S. fiscal deficit (Roubini and other, 2005; Roubini and Setser, 2004).

16

A review of the empirical literature on the income elasticity of U.S. imports by Marquez (2000), for instance, reveals that estimates for this coefficient have varied widely. Chinn (2004, 2005) finds evidence in favor of the Houthakker-Magee effect, although its magnitude diminishes when some components are excluded from imports.

17

See, for instance, Greenspan (2005) for a discussion of how changes in pass-through have influenced the current account.

18

The exchange rate is defined in units of foreign currency per U.S. dollar, so that a decline in the exchange rate corresponds to a depreciation. Pass-through coefficients are therefore expected to have a negative sign.

19

Marazzi and others (2005) have argued that commodity prices function as an indirect transmission channel for exchange rate changes. Indeed, including oil and commodity prices separately in equation (1) appears to weaken the estimated direct pass-through, even when restricted to prices of noncommodity goods imports. However, we found no strong evidence that in recent years, weaker pass-through may have been accompanied by stronger commodity price effects.

20

Interestingly, using import-weighted exchange rates leads to estimates of marginally weaker pass-through, possibly indicating that the “broad” effective exchange rate—which also incorporates information about third markets—captures more accurately the relevant relative price of the dollar.

21

The sample includes sectors with price data going back to at least 1987. The results are robust to extending the sample to include sectors with shorter price data series. See Annex for data sources.

22

Although this finding could simply indicate that China happened to specialize in products that exhibit smaller price fluctuation, it may also suggest that a large and increasing market presence by China may have a stabilizing effect on prices.

23

For sectors with shorter price data series, the pass-though coefficients for the earlier period are estimated from a correspondingly shorter sample.

24

Note that under this restriction, which has not been rejected by the data, both domestic and foreign income enter with the same sign, and both are afforded equal weight in determining the long-run income elasticity.

25

Indeed, the forecast errors increased slightly, further deepening the export puzzle.

26

It is also possible that the long-run sensitivity to relative price fluctuations have shifted. Unfortunately, the short sample is unlikely to provide enough information to study the stability of long-run relationships.

27

See for instance Kose and others (2004).

United States: Selected Issues
Author: International Monetary Fund
  • View in gallery

    Relative import prices have not risen as fast as the dollar has depreciated.

  • View in gallery

    Estimated short-term pass-through coefficients

    (all commodities)

  • View in gallery

    Estimated short-run pass-through coefficients

  • View in gallery

    U.S. Effective Exchange Rate Weights, 1995 and 2002

  • View in gallery

    China’s import share of selected end-use goods

    In percent of all U.S. imports by sector

  • View in gallery

    Import Prices and China’s Import Share

  • View in gallery

    Passthrough and China’s Share in U.S. Imports

  • View in gallery

    U.S. Trade Balance: In-sample forecasts and actual values

  • View in gallery

    Export and Import Volume Relative Price Elasticities

    (rolling 60 quarter correlation with +/- 1.65 standard deviation bands)

  • View in gallery

    U.S. Trade Balance: Out-of-sample forecasts